polmineR v0.8.9
split()
sets type of subcorpus
to
NA
, causing an error if another split is performed.
Fixed.
split()
throwed misleading error message if
s_attribute
not existing. The error message is now telling
#242.
split()
was not implemented if s_attribute was child.
Done #243.
- Inefficiency of
size()
for corpus
objects
for scenario of nested s-attributes addressed #231.
enrich()
for subcorpus_bundle
objects
(returning partition_bundle
now) #224.
subset()
implemented for subcorpus_bundle
obejcts #234.
- Sample corpus GERMAPARLMINI now includes s-attribute
protocol_lp.
- Bug removed from
setAs()
-method from slice
to “AnnotatedPlainTextDocument” that would prevent using GERMAPARLMINI
as sample data.
- Method
decode()
can return ‘AnnotatedPlainTextDocument’
from NLP package.
- Coerce method
as(x, "AnnotatedPlainTextDocument")
not
available any more.
- Method
decode()
has new argument “stoplist” to drop
terms from ‘AnnotatedPlainTextDocument’. Unused for other return
values.
- Tooltips now have auto width - all text is displayed.
- The s-attribute ‘role’ has been added to GERMAPARLMINI to make it
more suitable for demonstration purposes for data linkage.
- Improved documentation of method
get_template()
,
examples added.
- Formatting instructions for “subtitle” added to template file
“article.template.json” in folder “templates”.
- The
show()
-method for corpus
objects gives
an information whether a template is available.
- Bug removed from
as.markdown()
that would prevent
fulltext display for non-parliamentary-protocol documents.
- Method
tooltips()
has new argument fmt
to
provide flexibility to assign tooltips based on corpus positions.
- New function
href()
to add hypertext references to
fulltext output.
- Method
read()
has new argument annotation
to get values for arguments highlight
,
tooltips
and href
from a subcorpus
object.
- Internally, variants of opening/closing double quotes are removed
that interfere with html output.
- The
format()
method used internally to produce output
does not drop s-attributes ending on “_id” any more #253.
- The default value for argument
progress
is FALSE for
the hits()
method for character class objects, as a matter
of consistency #252.
- Performance improvement for application of values in
split()
for corpus
objects.
- The
decode()
method for subcorpus
objects
is now able to process nested corpora. Performance gain for all
scenarios.
as.TermDocumentMatrix()
for bundle
objects
speed ups instantiation of simple_triplet_matrix
.
- Method
s_attributes()
for bundle
objects
is implemented much more efficiently.
- Methods
get_token_stream()
and ngrams()
have new argument vocab
to pass in alternative dictionary.
Envisaged usage is to efficiently use pruned vocabulary for decoding the
token stream.
- New method
ngrams()
for list
objects.
Serves as worker for ngrams()
-method for
partition_bundle
objects.
- Better handling of conflicting registry directories by method
corpus()
.
- Method
get_token_stream()
for numeric
input has new argument registry
to optionally specify
registry directory.
- Method
count()
for subcorpus
objects did
not pass value of argument verbose
to cpos()
,
resulting in potentially unwanted verbosity. Fixed.
- Subsetting a
subcorpus
using
subset()
-method kept strucs for nested attributes but
assigned ancestor s-attribute to slot “s_attribute_strucs”, resulting in
false counts, for example. Fixed.
- The
split()
-method for subcorpus
objects
was not implemented correctly for descendent attributes without values,
so that getting subcorpora with sentences in a subcorpus would have
wrong result. Fixed.
- Argument
values
of method split()
for
corpus
objects did not process value FALSE
to
split corpus by s-attribute without values #263. Fixed.
- New
s_attributes()
-method for context
objects. Returns s-attribute values for the matches for query in context
object.
- Method
hits()
has new argument decoce
. If
FALSE
, the strucs for are not decoded.
s_attributes()
-method for expression
assigns types of vectors matched against as names if possible.
subset
for corpus
and subset
objects will use integer struc values for subsetting, if integer values
are passed in logical expression.
- Number of cores limited to 2 as required by CRAN Repository
Policy.
polmineR v0.8.8
- Improved performance of
enrich()
-method for
partition_bundle
objects #225.
- Refactored
as.TermDocumentMatrix()
for
partition_bundle
and bundle
, to improve
performance.
- Substantial performance improvement of
partition_bundle()
-method for partition
objects (more efficient instantiation of S4 objects).
- Performance improvement of
split()
-method for
subcorpus
objects.
- Defunct functions
store()
and mail()
have
finally been removed from the package.
- The
sample()
method for bundle
objects
(and objects inheriting from the bundle
class) did not yet
use the new convention to use single square brackets (not double
brackets) for extracting a subset from the bundle
. Fixed
#236.
- Performance improvements for
ngrams()
method for
partition_bundle
objects, introducing more efficient data
handling, vectorization and parallelization.
get_token_stream()
for partition_bundle
failed if all docs have equal length (mapply()
issue).
Fixed.
- Memory efficiency of
as.DocumentTermMatrtix()
for large
corpora significantly improved for handlung large corpora.
- The
$
-method for corpus
is now used for
accessing corpus properties, replacing previous usage to inspect
s-attributes.
- The
partition_bundle()
-method for context
class objects has improved verbosity now and telling progress
messages.
- New utility function
capitalize()
for uppercasing first
letter of elements in a character vector.
- The
trim()
-method for classes
DocumentTermMatrix
and TermDocumentMatrix
has
been updated. Arguments termsToKeep
, and
docsToDrop
have been deprecated, argument
termsToDrop
is deprecated and replaced by
terms_to_drop
and docsToKeep
is deprecated and
replaced by docs_to_keep
. New arguments
min_count
and min_doc_length
are introduced to
drop rare terms and short documents, respectively. The purpose of
redesigning the trim()
-method is to make it more useful for
preparing matrices for topic modelling.
- Method
subset()
for corpus
and
subcorpus
objects will now process indication of
s-attribute without value, so that subsetting corpora for s-attributes
without values is now possible.
- Method
split()
for subcorpus
objects will
now also work if s_attribute
for splitting is not a sibling
of the s-attribute the subcorpus is based on.
- Method
as.speeches()
for subcorpus
objects
refactored to work with nested scenario.
- Adapted to changes of pkg markdown >= 1.3 #235.
s_attributes()
will return NA
if
s-attribute does not have values #234.
hits()
-method for partition_bundle
objects
passes argument p_attribute
to cpos()
#239.
use()
returns TRUE
, if loading corpus in
package was successful, or FALSE
if not. Previously, the
function aborted with an error, or returned NULL
.
- If package ‘GermaParl2’ (with GERMAPARL2MINI inside) is available,
some initial tests for functionality for nested corpora is run.
- Subsetting a corpus using
subset()
would loose specific
subcorpus class (such as “plpr_subcorpus”). Fixed.
- Class “corpus” has slot “xml”, and classes (“subcorpus” and
“partition”) now inherit this slot.
html()
for subcorpus
reconstructs
meta
equivalent to read()
for
subcorpus
objects.
- Subsetting using an s-attribute without values now possible
#240.
polmineR v0.8.7
New features
- Using the
corpus
class throughout is an opportunity to
keep the corpus ID together with the registry directory of a corpus. And
as we are able now to handle corpora defined in different registry
files, the temporary registry directory is not necessary any more. It
still exists, yet only for temporary corpora and corpora that are
described by registry files that cannot be modified, i.e. corpora
shipped in packages. The test corpus of the polmineR package is an
important respective scenario.
get_token_stream()
now has an argument
min_length
.
registry_*()
functions are superseded by
RcppCWB::corpus_*
functions and throw a warning that they
are deprecated.
- The REUTERS corpus is not included in the package any more: There
was an identical copy of the REUTERS corpus included in the RcppCWB
package. All examples and unit tests now use
use(pkg = "RcppCWB", corpus = "REUTERS")
to make the
REUTERS corpus available.
size()
works for
partition
/subcorpus
with
s-attribute
that is a child of the s-attribute the object
is based on #216.
- The
trim()
-method for context
objects has
a new argument fn
for supplying a (trimming) function to be
applied all match contexts.
- A new s-attribute “protocol_date” has been added to sample corpus
“GERMAPARLMINI”, so that sample data for nested corpus data is
available. To prevent confusion between s-attributes “protocol_date” (at
protocol-level) and “date” (at speaker-level), argument
s_attribute_date
is stated explicitly in all examples.
- Method
size()
has been refactored to work with nested
corpora.
- Method
encoding()
and replace method
encoding<-
are defined for call
and
quosure
objects to get and adjust the encoding, replacing a
previously unexported function .recode_call()
.
- The
subset()
methods for corpus
and
subcorpus
objects now handle expressions for subsetting as
quosures, laying the ground to program against subset(), see respective
update of the examples, #212.
- Functionality for indexing
bundle
objects with single
square brackets is developed now. Indexing with double brackets,
suppling multiple values for i
is deprecated. The aim is a
consistent behavior that a bundle
indexed by [
will always return a bundle
, and indexing with
[[
always gets a single object from the list of objects.
#214
Minor improvements
- The
use()
function now has an additional argument
corpus
to specify which corpus from a package shall be
loaded (#138).
- The
get_token_stream()
-method for
partition_bundle
objects is more memory efficient (no
exhaustion for big corpora) and faster.
- Significantly improved performance of
split()
-method
for corpus
objects.
- The
split()
-method for corpus
objects
offers progress bar.
as.speeches()
for corpus
objects has new
argument subset
, offering a significantly faster approach
than the method for subcorpus
objects in many cases.
- The
size()
method will return NA
and issue
a telling warning if the slot corpus
and
registry_dir
of the corpus
object are not
filled #222.
get_token_stream()
will return list of
integer
values if decode
is TRUE
(#213).
- After applying
trim()
on a context
object
using arguments positivelist
or negativelist
,
the count
slot as reported by length
was not
updated. Fixed. (#220)
- The
enrich()
method for context
objects
has a new argument stat
for creating / updating the
data.table
in the slot stat
.
- Method
subset()
for subcorpus
objects has
been debugged to work with nested corpora.
- New option
polmineR.mdsub
configures substitutions that
are applied on markdown documents to prevent presence of characters that
would be misinterpreted as formatting instructions. Fixes #166.
- The messages issued by
check_cqp_query()
now include a
hint that argument check
can be used to omit checking the
CQP syntax to prevent false positives. Addresses #171.
Bug fixes
- The ability of
cooccurrences()
(and
context()
) to process more than one p-attribute has been
lost temporarily. Fixed. #208.
- Removed a bug for
hits()
method for
partition
objects #215.
- After applying
trim()
on a context
object
using arguments positivelist
or negativelist
,
the count statistics reported in the stat
slot were not
updated. Fixed. (#220)
- Structural attributes do not disappear any more after adding
tooltips to a
kwic
object #218.
- Method
subset()
would not work reliably with argument
regex
if more than one expression is passed #212.
Fixed.
terms()
did not work for subcorpus
objects. Fixed. #209
- When applying
as.speeches()
on a
subcorpus
, the date may have been missing from the object
names. Fixed. #219
- Fixed an issue that
minNchar
in the
noise()
method would work exactly the way opposite to the
way intended #211.
- The slot
registry_dir
of a
cooccurrences_bundle
derived from a
partition_bundle
was not filled, resulting in an error of
the show()
-method for the
cooccurrences_bundle
. Fixed #222.
Documentation
- The documentation of the
cooccurrences()
method now
includes example code for creating a table using
DT::datatable()
with buttons for exporting tables (to
Excel, for instance).
polmineR v0.8.6
New Features
- The
dispersion()
method now accepts an argument
fill
, a logical
value to explicitly control
whether (#160) zero matches for a value of a structural attribute should
be reported. The performance of adding columns (requred only if two
structural attributes are provided) is improved substantially by using
the reference semantic of the data.table package. If many columns are
added at once, a warning issued by the data.table package is
supplemented by an further explanatory warning of the polmineR package.
Filling up the data.table
was limited previously to
freq = FALSE
, this limitation is lifted.
- The
html()
method is implemented for
remote_subcorpus
objects.
- The
hits()
method is implemented for
remote_corpus
and remote_subcorpus
class
(#160).
- A new S4 class
ranges
is introduced to manage ranges of
corpus positions for query matches. This is a preparatory step to remove
an inconsistency from the hits
class that mixed two very
usages (getting ranges of corpus positions for matches and getting
counts).
- A new S4 method
ranges
serves as the constructor to
prepare a ranges
class object. In combination with
as.data.table()
, it replaces former functionality of
hits()
without argument s_attribute
.
- The output of the
hits()
method is altered, making it
much more consistent than previously: The method will consistently
return a hits
object.
- The method
hits()
has a new argument fill
that will report zeros for combinations of s-attributes with no matches
for a query.
- The argument
subset
for the subset
method
for remote_corpus
objects can now be a call (#162), this is
a basis for passing vectors to OpenCPU server. -
p_attributes()
implemented for remote_corpus
and remote_partition
.
- A new
regions()
method (for corpus
class
objects to start with) returns a regions
class object with
a regions matrix (slot cpos
) with regions for an
s-attribute (#176).
- The
get_token_stream()
-method for regions
and matrix
objects will now accept a logical argument
split
. If TRUE
, a list of character vectors is
returned. The envisaged use case is a fast decoding of sentences
(#176).
- A
encoding()
method has been defined if argument
object
is missing. Calling encoding()
will
return the session character set. If it cannot be determined using
localeToCharset()
, a UTF-8 session charset will be assumed.
Internally, encoding()
replaces a direct call of
localeToCharset()
to avoid errors that have occurred on
GitHub Actions with Ubuntu 20.04 (#188).
- If the session character set cannot be guessed by
localeToCharset()
(NA
return value), a startup
message will issue a warning that ‘UTF-8’ is assumed (#188).
- The
size()
method is now able to handle nested
s-attributes.
- The
trim()
method for context
objects will
now accept a matrix with ranges a positivelist
argument.
- The
highlight()
method now acceps matrix
objects as elements of the list of items to be highlighted. It is
treated as a set of regions, such as resulting from cpos()
.
Thus it is possible to highlight matches for CQP queries.
- The package now requires at least RcppCWB v0.5.2, which includes a
much more efficient worker for token contexts for the
context()
method.
- The
count()
-method for partition_bundle
objects failed with an opaque error message if there were no query
matches at all. There is now a check for this scenario and the expected
table is returned (zero values throughout.)
- The
corpus
class is now a superclass for the
textstat
class, starting to create a more coherent class
structure in general. This is an important preparatory step to be able
to keep all registry files in the temporary registry directory. To avoid
a confusion in the class system resulting from the coerce method from
partition
to corpus
objects, this coerce
method (defined by setAs()
) has been removed. The
get_template()
-method for partition
objects
using this coerce method has been removed - as it inherits the method
anyway, it is not needed any more. See #201.
- The kwic tab of the shiny app included in the package exposes the
improved capabilities to determine the context of a query match based on
an s-attribute (argument
region
) and to consider the
changing value of an s-attribute as a boundary of a context (argument
boundary
). New menu “boundary” and radio buttons,
conditional on presence of s-attributes “s” and/or “p”.
Minor Improvements
- If arguments
sAttribute
or pAttribute
(instead of s_attribute
and p_attribute
) are
still used with dispersion()
method, a warning is issued
declaring that the argument is deprecated.
- Examples in packages that depend on polmineR would have faced the
issue that loading/re-loading the package in several examples would not
be posssible as the mechanism of cleaning up between examples would
trigger a removal of polmineR’s temporary directories but not the
re-creation. Removing temporary files is now moved from polmineR’s
.onDetach()
to .onUnload()
(#164).
- Significant improvement of the performance of the
as.phrases()
method (#172).
- The
as.corpusEnc()
auxiliary function will now check
whether non-convertible characters lead to an NA
result and
issue a warning how this warning can be avoided (#151).
- Significant performance improvement of the
context()
method for matrix
objects if arguments left
and right
are named integer
vectors. All
context()
benefit from the improved performance of this
worker for creating contexts for query matches.
- New coerce-method to derive matrix with ranges from a
context
object.
- The
enrich()
method for context
objects
will now perform an in-place operation when adding new
s-attributes.
- The
as.cqp()
function includes arguments
check
and warn
for running
check_cqp_query()
on queries.
- The
context()
method for matrix
objects
includes a new argument boundary
and relies on a new
functionRcppCWB::region_matrix_context()
.
- Default value of argument
verbose
of
context()
-methods is now FALSE
.
- The
as.corpusEnc()
auxiliary function now includes a
test whether input character vector includes unexpected encodings and
issues a warning if this is the case.
- The
cpos()
method will now check for accidental leading
and/or trailing whitespace and remove it for token lookup. Note that
hits()
, count()
and dispersion()
will report queries without removing whitespace.
- Internals of the
count()
-method for
partition_bundle
objects will be much more efficient when
many columns with zero matches need to be added. The implementation
avoids a data.table warning when the bulk action of adding new columns
exceeds the number of columns reserved by data.table objects.
- The DESCRIPTION files does not state “LazyData: yes” any more, as
the package does not have a data directory.
- Typo in messages of
trim()
is removed (#197).
encoding()
relies on l10n_info()
before
using localeToCharset()
as a matter of performance and
robustness (#196).
- Class
corpus
has a new slot registry_dir
.
This is a preparatory step that will facilitate managing corpora
described by registry files in different registry directories.
- Constructor
corpus()
for corpus
-class
objects has an argument registry_dir
that will be required
to distinguish corpora described by registry files in different registry
directories.
- The package now relies on the the fs package to handle directories
and paths. Slots in S4 classes are not
fs_path
classes.
- Internally, functions
registry_get_home()
and
registry_get_encoding()
have been replaced by RcppCWB
functions cl_charset_name()
and
corpus_data_dir()
with equivalent result, but faster due to
immediate access to C representation of the corpus.
- The
corpus()
method will deduce the registry directory
from the C representation of the corpus if possible.
- An inefficiency in the implementation of
as.markdown()
has been removed, making fulltext display (using read()
or
html()
) much faster.
- Calling
corpus()
without any arguments now returns an
expanded data.frame
reporting all slots of the
corpus
class objects, skipping only the data directory of
the corpus.
- The
cpos()
method for matrix
objects that
turns a matrix with corpus positions into a vector of
integer
values now relies on a C-level implementation newly
included in the RcppCWB package, that is significantly faster than the
best possible implementation in R.
- The table generated by
kwic()
shows row numbers, which
is convenient when referring to specific rows (#184).
- The
as.cqp()
now checks whether argument
query
meets the expectation that it is a query (#191).
- The method
make_region_matrix()
, which has been used
internally only, has been removed.
RcppCWB::s_attr_regions()
replaces the functionality.
- The
as.speeches()
method had not yet been implemented
for nested corpora. A limited rewrite makes this work now (#198).
- Inconsistencies and unnecessary limitations of the
get_token_stream()
method for partition_bundle
objects have been addressed: Multiple p-attributes can be used without
providing phrases
at the same time (#142) and using the
subset
argument does not depend on using
phrases
either (#141).
- The
as.sparseMatrix()
method is now also defined for
DocumentTermMatrix
objects (was available previously ony
for TermDocumentMatrix
objects).
- If a vector of queries is named, theses named are now used
consistently by the
hits()
method (#195).
get_type()
for subcorpus_bundle
returns
NULL
if no type is defined as a matter of consistency
(#169).
- If an expression for subsetting a
corpus
/subcorpus
includes invalid
s-attributes, the warning is telling and NULL
is returend
(#179).
- The cooccurrences options of the shiny app mirror the arguments
used/required by the
cooccurrences()
method - left/right
rather than window (#134).
- Methods
kwic
and context
now have argument
region
as an intuitive alternative to named
character
vectors left
and right
when expanding match to left and right limitation of an
s-attribute.
Bug fixes
- A limitation to pass long arguments to an OpenCPU server resulting
from
deparse()
within is resolved (#161).
- The
hits()
method for the slice
virtual
class has been removed and the implementation for hits
for
the subcorpus
class is now real worker, also invoked for
hits()
for partition
. This removes a bug that
occurred when applying hits
on subcorpus
objects, which resulted in a count for the whole corpus.
- Shortcoming of the
show()
-method for
partition
objects resvolved when more than one s-attribute
has been used to define partition
(#170).
- Arguments
left
and right
of the
context()
-method for matrix
objects, the
worker behind the context()
, kwic()
and
cooccurrences()
methods did not work as intended for
character
values specifying an s-attribute. Fixed - it is
not possible to use these arguments (#173).
- An error that occurred with
as.TermDocumentMatrix()
or
as.DocumentTermMatrix()
when a s-attribute would not cover
the entire corpus has been removed (#177). In this vein, an efficiency
(decoding token stream twice) has been removed, so performance will also
be better.
- An error that occurred temporarily when passing an expression with
logical operators without substituting the expression to
subset()
for remote_corpus
objects(#181) has
been fixed.
- The
context()
method, and kwic()
for
partition
or subcorpus
objects did not process
left and right contexts correctly, if it was a named character vector.
Fixed.
- The
hits()
method failed for
partition_bundle
objects when there were no matches for the
query. Fixed. (#199 and #163)
- The
p_attributes()
method for slice
objects had an error when decoding the token stream. Fixed.
- An error when using
format()
on a
features_ngrams
object resulting in an error when using
knit_print()
on this object has been fixed (#200).
- The
edit()
method can now be invoked on a
features
object (#165).
- The
context()
-method for partition_bundle
objects always required an explicit statement of the argument
positivelist
, which is not necessary. Fixed. (#178)
- A bug reported for the progress bar of the
kwic()
method is gone as a result of refactoring how the s-attribute is matched
(#149). The argument progress
has been removed from the
method.
- The
as.DocumentTermMatrix()
method mistakenly returned
as TermDocumentMatrix
object. Fixed (#146).
- The
noise()
method misleadingly handled the number of
characters provided by minNchar
as a maximum threshold, not
as a minimum requirement (#135). Fixed.
Documentation
- Checks in examples whether magrittr is available have been dropped,
as magrittr has become a dependency and the pipe operator is available
by default.
- The documentation of the
hits
class now describes the
data.table
in the stat
slot of the class in
detail.
polmineR 0.8.5
New Features
- A new
decode()
method for data.table
objects shall serve as a more user-friendly access to the efficiency of
the RcppCWB::cl_cpos2str()
function.
- The
data.frame
returned when calling
corpus()
will now include a column with the encoding of the
corpus.
Bug fixes
- The
warn
argument of the
get_template()
-method remained unused, resulting in a
warning message even if warn
was FALSE
,
resulting in a set of warning messages when calling
corpus()
. The argument is used as intended now and defaults
to FALSE
.
- The
as.markdown()
-method for subcorpus
objects now uses an (internal) default template accessible via
polmineR:::default_template
, if no template is defined for
a corpus.
- The
registry_get_encoding()
function returned a
length-one character vector if the regular expression to extract the
charset corpus property did not yield a match. To prevent errors, it now
returns “latin1” as the CWB standard encoding (#159).
polmineR 0.8.4
Minor Improvements
- The
knit_print()
-method for textstat
objects does not accept the three dots argument any more. As an
installation of pandoc is necessary to include resulting
htmlwidget
in an html document, the method will check now
whether pandoc is available. If not, a formatted data.table
is returned.
- The
knit_print()
-method for kwic
objects
does not have the pagelength
argument any more as it has
been unused. The pagelength is controlled by the option
polmineR.pagelength
. Internally, the method will call the
method for the textstat
superclass of the kwic
class, which is newly robust against a missing installation of
pandoc.
- Any Unicode characters that could be detected have been removed from
the documentation to avoid warnings on the CRAN Solaris test machine
(#156).
Bug Fixes
- The
chisquare()
method needs to increase the number of
digits temporarily, but failed to revert to the original value as
expected. One implication was, that rounding the values in
data.table
objects would fail, and rounding in general
yielded very strange results (#155). Fixed.
polmineR 0.8.3
Minor Improvements
- The
as.data.table()
-method defined in the
data.table
is now reexported and defined and documented for
the textstat
, regions
and bundle
class that it can be used cleanly.
- The installation instructions have been removed from the package
vignette. The logical place for these instructions is the README.md file
and this will be single place where users will find authoritative
up-to-date installation instructions.
polmineR 0.8.2
Minor Improvements
- The (well-hidden)
.importPolMineCorpus()
-function has
been superseded by cwbtools::corpus_install()
and has been
removed from the package.
- Usage of
cat()
has been replaced by
massage()
within functions throughout to meet CRAN
requirements.
- The unused argument
type
has been dropped from the
html()
-method for partition_bundle
objects.
- The
html()
-method for character
class
objects now serves as a worker to generate html from markdown. The
html()
-method for partition_bundle
objects did
not return a html
class object as stated in the
documentation object. Fixed.
- The
store()
-method has been declared defunct as it is
unnecessary functionality that bloats the package. Using
format()
in combination with
openxlsx::write.xlsx()
is the recommended alternative
workflow.
- The
mail()
-method has been declared defunct and has
been removed from the package. A more user-friendly workflow is to use
export buttons of the DataTable widgets.
- The
Corpus
class has been removed from the package as
it has beeen defunct for a while.
- To avoid the side-effects of the
set_template()
method
on options that may be unnoticed for the user and that potentially
violate CRAN policies, the method has been dropped.
Bug Fixes
- The
s_attributes()
-method returned a
data.table
mixing up rows / columns for
subcorpora/partitions with a region matrix that would only include a
single set of corpus
polmineR 0.8.1
New Features
- The
decode()
-method now entails the possibility to
decode structural and positional attributes selectively, via new
arguments p_attributes
and s_attributes
(#116). Internally, the reliance on coerce()
-methods has
been replaced by a simpler if-else-syntax. The
as(from, "Annotation")
option persists, however.
- A new argument
phrases
was added to the
count()
-method for partition_bundle
objects.
- The slots “user” and “password” of the
remote_corpus
and the remote_subcorpus
class are replaced by a single
slot restricted
(values
TRUE
/FALSE
) to indicate if a user name and a
password are necessary to access a corpus. A file following the
conventions of CWB files is assumed to include the credentials for
corpus access. This approach avoids the accessibility of the
password.
- Using the temporary registry file can be suppressed by setting the
environment variable POLMINER_USE_TMP_REGISTRY as ‘false’. (Background:
Necessary to deal with changing temporary directories when polmineR is
preloaded in an OpenCPU context.)
- The Dockerfile included in the package
(./inst/docker/debian_polminer_min) prepares a Debian image with a
minimal installation of polmineR that will be available at the ‘polmine’
repository at dockerhub (see
https://hub.docker.com/r/polmine/debian_polminer_min
).
- The
corpus()
-method that serves as a constructor either
for the corpus
or the remote_corpus
class does
not flag default values for the arguments user
and
password
any more. If the argument server
is
stated explicitly (not NULL
, default), these variables will
get the value character()
. This way, a set of if/else
statements can be omitted and it is much easier to implement methods for
the remote_corpus
class for corpora that are
password-protected, or not.
- There is now a definition of an S3
as.list.bundle()
-method (previously, there has only been
the S4 method). The nice consequence is that lapply()
and
sapply()
can be used on bundle
objects now (a
subcorpus_bundle
, for instance)
- The performance of the
count()
-method for
partition_bundle
objects has been improved, it is twice as
fast now (#137).
- The
p_attributes
method now accepts an argument
decode
.
- The
p_attributes
-method has been implemented for
partition_bundle
objects.
- In the shiny app you can launch via
polmineR()
, the
mail-button has been dropped in the kwic, and code can be displayed
(using code highlighting)
- The settings have been dropped from the shiny app altogether, as we
have the buttons now the
phrases
argument is used are now
also available when a phrases
object is not passed in.
- Code buttons have been added to the shiny app experimentally.
- The
get_token_stream()
-method for
partition_bundle
objects will now accept an argument
phrases
(#128).
- The
merge()
-method for
partition_bundle
-objects has been reworked: Substantial
performance improvement by relying on
RcppCWB::get_region_matrix
. Internally, the method performs
a check whether the partition
/subcorpus
objects to be merged are non-overlapping. The default value for the
argument verbose
is now FALSE
, as waiting time
is much shorter.
Minor Improvements
- A new option
polmineR.warn.size
can be used to control
the issuing of warnings for large kwic
objects.
- Indexing
Cooccurrences
objects had not been possible,
now at least using integer indices is possible (#114).
- Introduced experimentally a feature to count phrases in the
count()
-method for slice
class objects.
- The
corpus()
method for a character vector will now
abort gracefully with a message if more than one corpus is offered as
.Object
.
- The
Cooccurrences()
-method will now accept zero values
(0) for the arguments left
and right
. Relevant
for detecting bigrams / phrases.
- When sorting the results
data.table
of a
Cooccurrences
object, the NA values are pushed to the end
of the table now.
- A new
concatenate()
method is a worker to collapse
tokens into phrases.
- Implemented pointwise mutual information (PMI) for
Cooccurrences
class objects, see
pmi()
-method.
- Implemented a
ngrams()
-method for class
data.table
- useful if you need to work with decoded
corpora.
- Implemented the
pmi()
-method for the
ngrams()
-method, to provide a workflow for phrase
detection.
- A new method
enrich()
for object of class
Cooccurrences
will add columns with counts for the
co-occurring tokens to the data.table
in the slot
‘stat’.
- Removed an inconsistency with the naming of the columns of the
data.table
in the stat
slot of an
ngrams
object: Column names will now be “word_1” , “word_2”
etc.
- Defined an explicit method
count()
for
subcorpus_bundle
objects (just callling
callNextMethod()
internally) - useful to see the
availability of the method in the documentation object.
- The
as.speeches()
-method for corpus
objects now supports parallelization
- A unit test checks different methods for generating a
DocumentTermMatrix
against each other, as a safeguard that
different approaches might lead to different results (#139).
- New class
phrases
and as.phrases()
-method
for ngrams
and matrix
objects. The
count()
-method now accepts an argument
phrases
. See the documentation
(?phrases
).
- The
s_attributes()
-method is now consistent with the
usage of the unique
argument (#133).
- The
hits()
-method for partition_bundle
objects now accepts an argument s_attribute
to include
metadata in results (#74).
- The
check_cqp_query()
function now has a further
argument warn
. If TRUE
(default), a warning is
issued, if the query is buggy. The as.phrases()
-method will
use the function to avoid that buggy CQP queries may be generated.
- If no template is set, no reliance on a plain and simple template,
and telling error messages, if no template is available (#123).
- The
Corpus
class has been re-introduced (temporarily),
to avoid an issue with the GermaParl package if the class is not
available (#127).
- The
get_template()
-method is now defined for the
corpus
class.
- The
count()
-method with arguments
breakdown
is TRUE
and cqp
is
TRUE
has been awfully slow. Fast now.
- Decoding a p-attribute has seen a substantial performance
improvement (#130). A new argument
boost
allows user to opt
for the improvement, which will involve decoding the lexicon
directly.
- The
merge()
-method is implemented for
subcorpus_bundle
objects now, and has been implemented for
subcorpus
objects (#76).
- Generating a
kwic
view from a
cooccurrences
object based on more than one p-attribute
will work now (#119).
- The
decode()
-method has been defined for
integer
vectors. Internally it will decide whether decoding
token ids is speeded up by reading in the lexicon file directly. The
behavior can be triggered explicitly by setting the argument
boost
as TRUE
.
- The
get_token_stream()
-method will use the new
decode()
-method for integer values internally. The argument
boost
is used by the get_token_stream()
to
control the approach.
- Improvements of performance initially implemented for
get_token_stream
for partition_bundle
.
- Internally, the
partition_bundle()
-methods defined for
character
, corpus
and partition
objects now call the split()
-methods for
corpus
and subcorpus
objects, resulting in a
huge performance gain (#112).
- Zero values can be processed by
Cooccurrences()
-method
(#117).
- The
corpus
class includes a (new) slot
size
, just as the regions
and the
subcorpus
classes.
- The
split()
-method for corpus
objects now
accepts the argument xml
, to indicate whether the
annotation structure of the corpus is flat or nested.
- The definition of the S4 class
partition
now includes a
prototype defining default values for the slots ‘stat’ (a
data.table
) and the slot ‘size’ (NA_integer_
).
This avoids that an incomplete initialization of a
partition
object will result in an error.
- The
kwic()
-method is now available for
partition_bundle
/subcorpus_bundle
-objects
(#73).
- To make the
kwic()
-method work correctly for
partition
objects that result from a merge()
operation, the cpos()
-method for slice
objects
will extract strucs based on the s-attribute defined in the slot
s_attr_strucs
rather than the last s-attribute in the list
of the slot s-attributes
.
- Class
subcorpus
is exported for usage in other
packages.
- The default value of the argument
progress
of the
count()
-method for partition_bundle
objects is
now FALSE.
- The
get_type()
-method is now defined for the
corpus
class.
- Upon starting the shiny app included in the package, the presence of
packages “shiny” and “shinythemes” is checked. If the packages are not
yet present, an optional install is offered (#110).
- A coerce method has been defined to turn a
corpus
object into a subcorpus
object, to recover functionality
used (internally) that relied on the former Corpus
reference class.
- The
Cooccurrences()
-method is now defined for the
corpus
-class, too. The Cooccurrences()
-method
for the character
class now relies on this method.
- The deprecated
Corpus
reference class has been dropped
from the code altogether: As roxygen::roxygenize()
started
to check the documentation of R6 classes and reference classes, the poor
documentation of this class started to provoke many errors. Rather than
starting to write documentation for a deprecated class, getting rid of
an outdated and poorly documented class appeared to be the better
solution.
- New coerce method to derive a
kwic
object from a
cooccurrences
object. Introduced to serve as a basis for
quantitative/qualitative workflows, e.g. integrated in a
flexdashboard.
- There is now a telling error message for the
s_attributes()
method for corpus
objects when
values are requested for an s-attribute that does not exist (#122).
- In the
decode()
-method for subcorpus
objects, s-attributes were not decoded appropriately (#120). Fixed. When
decoding a corpus/subcorpus, the struc column is kept (again).
- A new check in
.onLoad()
whether polmineR is loaded
from the repository directory will ensure that temporary registry files
will not be gone when calling devtools::document()
(#68).
Bug Fixes
- In the
as.speeches()
-method for corpus
objects, setting progress
as FALSE
did not
suppress the display of a progress bar. Solved.
- Removed a bug that occurred when counting matches for CQP queries
over a
subcorpus_bundle
that resulted from CQP queries
being turned into invalid column names.
- Solved: No longer an error when calling polmineR commands after
having worked in the shiny app context (#111).
- A bug caused when the name of an object in a
partition_bundle
was an empty string and calling
count()
on this object has been removed (#121).
- A bug was addressed that occurs when unfolding the region matrix
where all regions have the same length (#124).
Documentation
- A skeleton documentation of package options is included in the
documentation of the package as a whole (
?polmineR
)
polmineR 0.8.0
New Features
- The
corpus
class has been put in a shape to become the
default point of departure of most workflows. All core methods are now
available for the corpus
class, and have been implemented
newly if necessary, e.g. show()
and
size()
-method. The constructor method for a
corpus
object, the corpus()
method, will now
check whether the character vector with the corpus ID refers to an
available corpus, whether all letters are upper case and issue
informative warnings and error messages.
- The
s_attributes()
-method for corpus
objects has been reworked: It will decode binary files directly, without
reliance on the corpus library functions, which is significantly
faster.
- The
Corpus
reference class is now obsolete after the
introduction of the S4 corpus
class. To maintain the
functionality not covered otherwise, new generics get_info
and show_info
have been introduced and defined for the
corpus
class.
- Methods available for the
subcorpus
class have been
expanded so that this class can supersede the partition
class: Methods newly available are cpos()
,
count()
, p_attributes()
,
s_attributes()
get_token_stream()
, and
size()
. Technically, there is virtual
slice
-class, from which subcorpus
inherits
(methods called via callNextMethod()
).
- A new
subset()
-method for the corpus
and
subcorpus
classes to generate subcorpora
(i.e. subcorpus
objects) has been introduced. It
outperforms the partition()
method. The
subset()
-method for corpus
and
subcorpus
objects will be the default way to work with non
standard evaluation in a manner that feels “R-ish” (#40).
- The
zoom()
-method that has been introduced
experimentally has been dropped again in favor of the
subset()
-method to get subcorpus
objects from
corpus
and subcorpus
objects. A set of
experimental methods for an initial check of the feasibility of a
non-standard evaluation approach to the generation of subcorpora has
been dropped (methods $
, ==
, !=
,
zoom
for corpus
-class).
- To facilitate the transition from the
partition
class
(inheriting from the textstat
class) to the
subcorpus
class (inheriting from the textstat
class), there is a new coerce()
-method to turn a
partition
object into a subcorpus
object.
- A new
remote_corpus
-class is the basis for accessing
remote corpora. A remote_subcorpus
can be derived from a
remote_corpus
. Methods available for remote corpora und
subcorpora remain limited at this stage.
- Consolidation of the class system: For all the S4 classes in the
package, multiple contains have been checked, and multiple contains have
been removed.
- The
subcorpus_bundle
class now inherits from
partition_bundle
. This is not intended to be a long-term
solution, but facilitates the implementation of new workflows based on
the subcorpus
class rather than the partition
class.
- Calling the polmineR shiny app via
polmineR
did not
have safeguards if the suggested packages shiny and shinythemes
were not installed. Now there will be a conditional installation of the
packages required for running the shiny app.
- The somewhat odd class
CorpusOrSubcorpus
has been
removed. The ngrams
-method now applies for
corpus
and subcorpus
objects.
- The pipe operator of the magrittr package
is imported now, and magrittr has moved from a suggested package to a
required package.
- The
label()
-method, present for a while, is superseded
by a edit()
-method now. It will call a shiny gadget either
using DataTables or Handsontable. The former Labels
reference class has been turned into a S4 class, because the desired
reference logic can also be achieved with a data.table
in a
slot of the labels class.
- The
table
-slot of the kwic
class has been
renamed as stat
slot (a data.table
), so that
the kwic
class can now inherit from the
textstat
class. The enrich()
-method for
objects of class kwic
now includes a new argument
extra
that will add extra tokens to the left of the windows
for concordances so that qualitative inspections for query hits can work
with more context.
- The
as.TermDocumentMatrix()
and the
as.DocumentTermMatrix()
-methods are now also defined for
kwic
objects. They work exactly the same as for the
context
class. To avoid having to write new methods, a new
neighborhood
virtual class has been introduced. The
aforementioned methods are defined for the virtual class and are
available for context and kwic class objects.
- Added CQP functionality to count tab in shiny app, and to the
dispersion tab.
- There is now a basic implementation of
get_token_stream()
for a partition_bundle
object.
- The
Cooccurrences()
-method is now available for
subcorpus
-objects (#88).
- There is a new coerce method to turn a
kwic
-object into
a context
-object. The neighborhood
virtual
class could be discarded again, and a bug could be removed that left an
enrich()
-operation for kwic
objects (argument
p_attribute
) ineffectual (#103).
- A potential error resulting from setting argument
cpos
to FALSE
in the kwic()
-method has been solved
(#106), and the documentation of the argument has been rewritten so that
includes a warning to use the argument falsely.
- If the properties “version” and “build_date” are available in a
registry file, the information will be shown when calling
use()
(#72).
Minor changes
- Added a new argument
regex
to the
cpos()
-method (for corpus
objects), which will
interpret argument query
as a regular expression. This may
be faster than taking query
as an outright CQP query.
- The configure-script in the package that would adjust paths in the
registry files for the corpora included in the package for documentation
and testing purposes has been removed. Having switched to a temporary
registry directory, it has lost its function.
- The version of the data.table package now required is 1.12.2,
because previous versions did not allow adding columns to a new
data.table.
- Implemented the possibility to use multiple queries in
dispersion
-method (#92).
- To keep up with the renaming of functions and arguments in the
package, “sAttributes” and “pAttributes” in the polmineR shiny app have
been renamed (“s_attributes”, and “p_attributes”, respectively).
- The shiny app module for kwic output will not show
p_attribute
and positivelist
by default.
- The
format()
-method is used to create proper output in
the cooccurrences of the shiny app.
- User names that include non-ASCII characters were a persistent
problem on Windows machines (#66). The solution now is to check for
non-ASCII characters in the path to the data directory, and to use the
“old” short DOS path if necessary. The worker is a modified
registry()
-function.
- The ordering of the table for
ll
-method had been
somewhat mixed up, which is repaired now. Tokens with NA values for the
ll-test will show up at the end of the table.
- The
registry_move()
-function, used only internally at
this stage, is exported now so that it can be used by other
packages.
- The return value of
the get_token_stream()
-method for
regions
objects was a data.table
. The behavior
is now in line with the other get_token_stream()
methods
- The
tempcorpus()
-method and the tempcorpus
class have been removed from the package, having become utterly
deprecated.
- The
summary()
-method for partition
-class
objects has been turned into a method for the count
-class,
to eliminate an inconsistency. The example of a workflow has been moved
to the documentation object for the count
-class.
- The
browse()
-method has not proven to be useful and has
been removed from the package. A new browse()
-function is
introduced to throw a warning, if browse should be called
nevertheless.
- A refactoring of the
split()
-method for
partition
-objects improved the readability of the code, but
the performance gain is minimal.
- A new
kwic_bundle
-class has been introduced, a list of
kwic
objects can be turned into this new class using
as.bundle
.
- The
context()
-method will now take again as input
character vectors for the arguments left
and
right
to expand to the left and right boundaries of the
designated region (#87).
- Rework of the way messages are printed to make it easy to implement
notifications in the shiny environment.
- Default highlighting when a positivelist is supplied has been
removed from the
kwic()
-method. This ensures that
subsequent highlighting operations can assign new colors (#38).
- Implemented feature request for
dispersion()
that
results are reported for all values of structural attributes, including
those with zero matches. (#104)
- Performance improved for the
cpos
-method for
matrix
which unfolds a matrix with regions of corpus
positions, useful for operations that require many calls.
- The
count
-method for partition_bundle
has
been reworked and is much faster and more memory efficient.
as.TermDocumentMatrix()
for
partition_bundle
optimized to work efficiently with large
corpora.
- Introduction of a context,matrix-method to have a unified auxiliary
function to create contexts.
- The
as.corpusEnc()
-function uses the
localeToCharset()
-function from the utils package to
determine the charset of input strings. On RStudio Server, we have seen
cases when the return value is NA. Then it will be assumed that the
locale is UTF-8.
- Functionality to highlight terms in kwic display has been restored
for the shiny app.
Bug fixes
- Removed a bug in the
context()
/kwic()
method that led to superfluous words in the right context.
- Removed a bug that occurred with the
as.data.frame()
-method for kwic
-objects when
no metadata were added.
- The
count()
-method for
partition_bundle
-objects did not perform
iconv()
if necessary - this has been corrected.
- Indexing the concordances of a
kwic
object did not
reduce the cpos
table concurringly. This has been
corrected.
- The
as.speeches()
-method failed to handle situations
correctly, when one speaker occurring in the corpus only contributed one
single region to the entire corpus (#86). This behavior has been
debugged.
- Counting over a
partition_bundle
started to throw a
warning that an argument arrives at the cpos()
-method that
is not used. The cause for the warning message is removed, an additional
unit test has been introduced to recognize issues with the
count
-method (#90).
- The
kwic()
-method threw an error when trimming the
matches by using a positivelist or a stoplist resulted in no remaining
matches. The method will now return a NULL object and keep issuing a
warning if no matches remain after filtering (#91).
- Chaining subsetting calls on a corpus/subcorpus omitted filling the
s_attribute slot of the
subcorpus
object, resulting in
false results when counting over subcorpora. Fixed.
- Started to remove bugs in the shiny app: kwic starts to work again
(bug: slot table has been replaced by stat).
- The part of the shiny app for dispersions did not work at all - has
been repaired, exposing more functionality of
dispersion()
(#62).
- In the
as.speeches()
-method, the argument
verbose
was not used (#64) - this had been addressed when
solving issue #86.
- Telling messages when sending out emails - on success and error -
have been added (#61).
- A shortcoming in coerce method to turn a
subcorpus
into
a String
was removed: A semicolon was not recognized as a
punctuation mark. This makes decoding subcorpora as
Annotation
more robust. The respective unit test has been
updated.
- Calling
read()
on a kwic
object works
again (#84).
- Checks for the
as.VCorpus()
method that failed are now
ok (#77). The reason was that get_token_stream()
assumed
implicitly that a p-attribute “pos” is present, which is not the case
for the REUTERS test corpus.
- A minor bug in the
s_attributes
-method was removed that
would make retrieving the metadata for the first strucs (index 0) of a
s-attribute impossible.
- Fixed an issue for
as.DocumentTermMatrix
that started
to occur with the introduction of the subcorpus_bundle
class (#100).
- Removed a bug in the
kwic
-method for
character
that prevented using different values for right
and left context (#101).
- Removed a bug that occurred when using
as.DocumentTermMatrix()
on a corpus stated by corpus ID /
length-one character vector (#105).
- Removed a bug from the kwic,character-method, and the
context,corpus-method that would result in odd behavior when either the
left or right context is 0.
- An endemic encoding issue for full text output on Windows machines
(latin1 encoding) has been solved by replacing internally
markdown::markdownToHTML
by a direct call to
markdown::renderMarkdown
. On this occasion, some overhead
preparing fulltext output has been removed.
- A bug that prevented getting extra left and right context for
kwic
objects has been removed (#102).
- The
as.TermDocumentMatrix()
-method for
neighborhood
-objects returned a DocumentTermMatrix
(unintendedly), this bug is removed now.
Documentation
- Extended documentation for
pmi()
-method and
t_test()
-method.
- New
s_attributes()
-method for
corpus
-class.
- The documentation for the
corpus
-class has been
rewritten entirely, and the documentation for the
remote_corpus
-class has been integrated, whereas methods
applicable to the remote_corpous
-class were integrated into
the documentation objects for the respective methods.
- The documentation for the
get_token_stream()
-method has
been reworked and expanded thoroughly (#65). On this occasion, test
coverage for the method has been improved significantly. (Everything is
tested now apart from parallelization.)
polmineR 0.7.11
NEW FEATURES
- A
Cooccurrences()
-method and a
Cooccurrences
-class have been migrated from the
(experimental) polmineR.graph package to polmineR to generate and manage
all cooccurrences in a corpus/partition
. A
cooccurrenes()
-method produces a subset of
Cooccurrences
-class object and is the basis for ensuring
that results are identical.
- New functionality to make using corpora more robust when paths
include special characters: There is now a temporary data directory
which is a subdirectory of the per-session temporary directory. A new
function
data_dir()
will return this temporary data
directory. The use()
-function will now check for non-ASCII
characters in the path to binary corpus data and move the corpus data to
the temporary data directory (a subdirectory of the directory returned
by data_dir()
), if necessary. An argument tmp
added to use()
will force using a temporary directory. The
temporary files are removed when the package is detached.
- Experimental functionality for a non-standard evaluation approach to
create subcorpora via a
zoom()
-method. See documentation
for (new) corpus
-class (?"corpus-class"
) and
extended documentation for partition
-class
(?"partition-class"
). A new corpus()
-method
for character vector serves as a constructor. This is a beginning of
somewhat re-arranging the class structure: The
regions
-class now inherits from the new
corpus
-class, and a new subcorpus
-class
inherits from the regions
-class.
- A new function
check_cqp_query()
offers a preliminary
check whether a CQP query may be faulty. It is used by the
cpos()
-method, if the new argument check
is
TRUE. All higher-level functions calling cpos()
also
include this new argument. Faulty queries may still cause a crash of the
R session, but the most common source is prevent now, hopefully.
- A
format()
-method is defined for textstat
,
cooccurrences
, and features
, moving the
formatting of tables out of the view()
, and
print()
-methods. This will be useful when including tables
in R Markdown documents.
- The
highlight()
-method for character
and
html
objects now has the arguments regex
and
perl
, so that regular expressions can be used for
highlighting (#99).
- The
as.data.frame()
-method for
kwic
-objects has seen a small performance improvement, and
is more robust now if the order of columns changes unexpectedly.
MINOR IMPROVEMENTS
- Startup messages reporting the package version of polmineR and the
registry path are omitted now.
- The functions
registry()
and data_dir()
now accept an argument pkg
. The functions will return the
path to the registry directory / the data directory within a package, if
the argument is used.
- The
data.table
-package used to be imported entirely,
now the package is imported selectively. To avoid namespace conflicts,
the former S4 method as.data.table()
is now a S3 method.
Warnings appearing if the data.table
package is loaded
after polmineR are now omitted.
- The
coerce()
-methodes to turn textstat
,
cooccurrences
, features
and kwic
objects into htmlwidgets now set a pageLength
.
- New methods for
partition_bundle
objects:
[[<-
, $
, $<-
- Rework of indexing
textstat
objects.
- A slot
p_attribute
has been added to the
kwic
-class; kwic()
-methods and methods to
process kwic
-objects are now able to use the attribute thus
indicated, and not just the p-attribute “word”.
- A new
size()
-method for context
-objects
will return the size of the corpus of interest (coi) and the reference
corpus (ref).
- New
encoding()
-method for character vector.
- New
name()
-method for character vector.
- A new
count()
-method for context
-objects
will return the data.table
in the stat
-slot
with the counts for the tokens in the window.
- The
decode()
-function replaces a
decode()
-method and can be applied to partitions. The
return value is a data.table
which can be coerced to a
tibble
, serving as an interface to tidytext (#37).
- The
ngrams()
-method will work for corpora, and a new
show()
-method for textstat
-object generates a
proper output (#27).
BUG FIXES
- Any usage of
tempdir()
is wrapped into normalizePath(…,
winslash = “/”), to avoid mixture of file separators in a path, which
may cause problems on Windows systems.
- In the calculation of cooccurrences, the node has previously been
included in the window size. This has been corrected.
- The
kwic()
-method for corpora returned one surplus
token to the left and to the right of the query. The excess tokens are
not removed.
- The object returned by the
kwic()
-method for
character
-objects method did not include the correct
position of matches in the cpos
slot. Corrected.
- Bug removed that occurrs when context window reaches beyond
beginning or end of a corpus (#48).
- When generating a
partition_bundle
using the
as.speeches()
-method, an error could occur when an empty
partition has been generated accidentaly. Has been removed. (#50)
- The
as.VCorpus()
-method is not available if the
tm
-package has been loaded previously. A coerce method
(as(OBJECT, "VCorpus")) solves the issue. The
as.VCorpus()`-method
is still around, but serves as a wrapper for the formal coerce-method
(#55).
- The argument
verbose
as used by the
use()
-method did not have any effect. Now, messages are not
reported as would be expected, if verbose
is
FALSE
. On this occasion, we took care that corpora that are
activated are now reported in capital letters, which is consistent with
the uppercase logic you need to follow when using corpora. (#47)
- A new check prevents an error that has occurred when a token queried
by the
context()
-method would occurr at the very beginning
or very end of a corpus and the window would transgress the beginning /
end of the corpus without being checked (#44).
- The
as.speeches()
-function caused an error when the
type of the partition was not defined. Solved (#57).
- To deal with issues resulting from an unset locale, there is a check
during startup whether the locale is unset (i.e. ‘C’) (#39).
- There was a difficulty to generate a
TermDocumentMatrix
from a partition_bundle
if the partitions in the
partition_bundle
were not named. The fix is to assign
integer numbers as names to the partitions (#58).
DOCUMENTATION FIXES
- Substantial rework of the documentation of the
ll()
,
and chisquare()
-methods to make the statistical procedure
used transparent.
- Expanded documentation for
cooccurrences()
-method to
explain subsetting results vs applying positivelist/negativelist
(#28).
- Wrote some documentation for the
round()
-method for
textstat
-objects that will show up in documentation of
textstat
class.
- Improved documentation of the
mail()
-method (#31).
- In the examples for the
decode()
-function, using the
REUTERS corpus replaces the usage of the GERMAPARLMINI corpus, to reduce
time consumed when checking the package.
polmineR 0.7.10
NEW FEATURES
- The package now offers a simplified and seamless workflow for
dictionary-based sentiment analysis: The
weigh()
-method has
been implemented for the classes count
and
count_bundle
. Via inheritance, it will also be available
for the partition
- and
partition_bundle
-classes. Then, a new
summary()
-method for partition
-class objects
is introduced. If the object has been weighed, the list that is returned
will include a report on weights. There is an example that explains the
workflow.
- The
partition_bundle
-method for
context
-objects has been reworked entirely (and is working
again); a new partition
-method for
context
-objects has been introduced. Buth steps are
intended for workflows for dictionary-based sentiment analysis.
- The
highlight()
-method is now implemented for class
kwic
. You can highlight words in the neighborhood of a node
that are part of a dictionaty.
- A new
knit_print()
-method for textstat
-
and kwic
-objects offers a seamless inclusion of analyses in
Rmarkdown documents.
- A
coerce()
-method to turn a kwic
-object
into a htmlwidget has been singled out from the
show()
-method for kwic
-objects. Now it is
possible to generate a htmlwidget from a kwic object, and to include the
widget into a Rmarkdown document.
- A new
coerce()
-method to turn
textstat
-objects into an htmlwidget (DataTable), very
useful for Rmarkdown documents such as slides.
- A new argument height for the
html()
-method will allow
to define a scroll box. Useful to embed a fulltext output to a Rmarkdown
document.
MINOR IMPROVEMENTS
- The
partition_bundle
-class, rather than inheriting from
bundle
-class directly, will now inherit from the
count_bundle
-class
- The
use()
-function is limited now to activating the
corpus in data packages. Having introduced the session registry,
switching registry directories is not needed any more.
- The
as.regions()
-function has been turned into a
as.regions()
-method to have a more generic tool.
- Some refactoring of the
context
-method, so that full
use of data.table
speeds up things.
- The
highlight()
-method allows definitions of terms to
be highlighted to be passed in via three dots (…); no explicit list
necessary.
- A new
as.character()
-method for kwic-class objects is
introduced.
BUG FIXES
- The
size_coi
-slot (coi for corpus of interest) of the
context
-object included the node; the node (i.e. matches
for queries) is excluded now from the count of size_coi.
- When calling
use()
, the registry directory is reset for
CQP, so that the corpora in the package that have been activated can be
used with CQP syntax.
- The script configure.win has been removed so that installation works
on Windows without an installation of Rtools.
- Bug removed from
s_attributes()
-method for
partition
-objects: “fast track” was activated without
preconditions.
- Bug removed that would swallow metadata/s-attributes to be displayed
in
kwic
-output after highlighting.
- As a matter of consistency, the argument
meta
has been
renamed to s_attributes
for the kwic()
-method
for context
-objects, and for the
enrich()
-method for kwic
-objects.
- To avoid confusion (with argument s_attributes), the argument
s_attribute
to check for integrity within a struc has been
renamed into boundary
.
- A new vignette “encodings” (rudimentary at this stage) explains what
users need to know about encodings when working with polmineR.
DOCUMENTATION FIXES
- Documentation for
kwic
-objects has been reworked
thoroughly.
polmineR 0.7.9
- new as.list,bundle-method for convenience, to access slot
objects
- as.bundle is more generic now, so that any kind of object can be
coerced to a bundle now
- as.speeches-method turned into function that allows partition and
corpus as input
- is.partition-function introduced
- sAttributes,partition-method in line with RcppCWB requirements (no
negative values of strucs)
- count repaired for muliple p-attributes
- bug removed causing a crash for as.markdown-method when cutoff is
larger than number of tokens
- polmineR will now work with a temporary registry in the temporary
session directory
- a (new) registry_move() function is used to copy files to the tmp
registry
- the (new) registry() function will get the temporary registry
directory
- the use() function will add the registry file of a package to the
tmp registry
- a bug removed that has prevented the name<- method to work
properly for bundle objects
- new partition_bundle,partition_bundle-method introduced
- naming of methods and functions, classes and most arguments moved to
snake_case, maintaining backwards compatibility
- utility function getObjects not exported any more
- for count,partition_bundle-method, column ‘partition’ will be a
character vector now (not factor)
- new argument ‘type’ added to partition_bundle
- new method ‘get_type’ introduced to make getting corpus type more
robust
- bug removed that has caused a crash when cutoff is larger than
number of tokens in a partition when calling get_token_stream
- count-method will now return count-object if query is NULL, making
it easier to write pipes
polmineR 0.7.8
- upon loading the package, check that data directories are set
correctly in registry files to make sure that sample data in
pre-compiled packages can be used
- startup messages adjusted slightly
polmineR 0.7.7
- removed depracated classes: dispersion, Textstat (reference class),
Partition (reference class)
- divide-methode moved to package polmineR.misc
- bug removed: size of ngrams object was always 1
- dotplot-method added for featuresNgrams
- sample corpus GermaParlMini added to the package (replacing
suggested package polmineR.sampleCorpus)
- configuration mechanism added to set path to data directory in
registry file upon installation
- class hits now inherits from class ‘textstat’, exposing a set of
generic functions (such as dim, nrow etc.); slot ‘dt’ changed to ‘stat’
for this purpose
- count,partitionBundle and hits,partitionBundle: cqp parameter
added
- RegistryFile class replaced by a set of leightweight-functions
(corpus_…)
- encode-method moved to cwbtools package
- getTerms,character-method and terms,partition-method merged
- examples using EUROPARL corpus have been replaced by REUTERS corpus
(including vignette)
- param id2str has been renamed to decode in all functions to avoid
unwanted behavior
- robust indexing of bundle objects for subsetting
- optional settings have been cleaned
- reliance on cwb command line tools removed
- encoding issue with names of partitionBundle solved
polmineR 0.7.6
- functionality of matches-method (breakdown of frequencies of
matches) integrated into count-method (new param breakdown)
- corpus REUTERS included (as data for testsuite)
- adjust data directory of REUTERS corpus upon loading package
- a pkgdown-generated website is included in the docs directory
- consistent use of .message helper function to make shiny app
work
- bug removed for count-method when options(“polmineR.cwb-lexdecode”)
is TRUE and options(“polmineR.Rcpp”) is FALSE
- if CORPUS_REGISTRY is not defined, the registry directory in the
package will be used, making REUTERS corpus available
- getSettings-function removed, was not sufficiently useful, and was
superseded by template mechanism
- new class ‘count’ introduced to organize results from count
operations
- at startup, default template is assigned for corpora without
explicitly defined templates to make read() work in a basic fashion
- new cpos,hits-method to support highlight method
- tooltips-method to reorder functionality of
html/highlight/tooltip-methods
- param charoffset added to html-method
- coerce-method from partition to json and vice versa, potentially
useful for storing partitions
- sAttributes2cpos to work properly with nested xml
- partition,partition-method reworked to work properly with nested
XML
- encoding of return value of sAttributes will be locale
- references added to methods count, kwic, cooccurrences,
features.
- as.DocumentTermMatrix,character-method reworked to allow for
subsetting and divergence of strucs and struc_str
- html,partition-method has new option beautify, to remove whitespace
before interpunctuation
- output error removed in html,partition-method (that misinterprets ``
as code block)
- the class Corpus now has a slot sAttribute to keep/manage a
data.table with corpus positions and struc values, and there is a new
partition,Corpus-method. In compbination, it will be a lot faster to
derive a partition, particularly if you need to do that repeatedly
- a new function install.cwb() provides a convenient way to install
CWB in the package
- added a missing encoding conversion for the count method
polmineR 0.7.5
- class ‘Regions’ renamed to class ‘regions’ as a matter of
consistency
- data type of slot cpos of class ‘regions’ is a matrix now
- rework and improved documentation for decode- and
encode-methods
- new functions copy.corpus and rename.corpus
- as.DocumentTermMatrix-method checks for strucs with value -1
- improved as.speeches-method: reordering of speeches, default
values
- blapply-method: verbose output will be suppressed of progress is
TRUE
polmineR 0.7.4
- applying stoplists and positivelists working again for
context-method
- matches-method to learn about matches for CQP queries replacing
frequencies-method
- Rework of enrich-method, including documentation.
- param ‘neighbor’ dropped from kwic,context-method; params
positivelist and negativelist offer equivalent functionality
- highlight-method for (newly exported) kwic-method (for validation
purposes)
- performance improvement for partitionBundle,character-method
- a new Labels class and label method for generating test data
- bug removed for partitionBundle,character-class, and performance
improved
- Improved explanation of the installation procedure for Mac in the
package vignette
- for context-method: param sAttribute working again to check
boundaries of match regions
- sample-method for objects of class kwic and context
- kwic, cpos, and context method will accept queries of length >
1
- use-function and resetRegistry-function reworked
- more explicit startup message to get info about version, registry
and interface
- encoding issues solved for size-method, hits-method and
dispersion-method
- use-function will now work for users working with polmineR.Rcpp as
interface
polmineR 0.7.3
- new installed.corpora() convenience function to list all data
packages with corpora
- view-method and show-method for cooccurrences-objects now
successfully redirect output to RStudio viewer
- data.table-style indexing of objects inheriting from
textstat-class
- for windows compatibility, as.corpusEnc/as.nativeEnc for encoding
conversion
- performance gain for size-method by using polmineR.Rcpp
- dissect-method dropped (replaced by size)
- improved documentation of size-method
- labels for cooccurrences-output
- cooccurrencesBundle-class and cooccurrence-method for bundle
restored
- as.data.table for cooccurrencesBundle-class
- count-method for whole corpus for pAttribute > 1
- functionality of meta-method merged into sAttributes-method
(meta-method dropped)
- speed improvements for generating html output for reading
- previously unexported highlight-method now exported, and more robust
than before (using xml2)
- progress bars for multicore operations now generated by pbapply
package
- starting to use testthat for unit testing
polmineR 0.7.2
- updated documentation of partition-method.
- documentation of hits-method improved
- use-methode: default value for pkg ist NULL (return to default
registry), function more robust
- Rework for parsing the registry
- rework of templates, are part of options now (see ?setTemplate,
?getTemplate)
- experimental use of polmineR.Rcpp-package for fast counts for whole
corpus
- new convenience function install.corpus to install CWB corpus
wrapped into R data package
- adjustments to make package compatible with polmineR.shiny
- cpos-method to get hits more robust if there are not matches for
string
- hits-method removes NAs
- compare-method renamed to features-method
- warnings caused by startup on windows removed
polmineR 0.7.1
- size-method now allows for a param ‘sAttribute’
- hits-method reworked, allows for names query vectors
- first version that can be installed on windows
polmineR 0.7.0
- rcqp package moved to suggests, to facilitate installation
polmineR 0.6.3
- more generic implementation of as.markdown-method to prepare use of
templates
- LICENSE file updated
- getTokenStream,character-method: new default behavior for params
left and right
- use of templates for as.markdown-method
- Regions and TokenStream class (not for frontend use, so far)
- getTermFrequencies-method merged into count-method
- Corpus class introduced
- decode- and encode-methods introduced
polmineR 0.6.2
- refactoring of context-method to prepare more consistent usage
- progress bar for context-method (using blapply)
- progress bar for partitionBundle (using blapply)
- more coherent naming of parameters in partitionBundle-method
- partitionBundle,character-method debugged and more robust
- usage of blapply in as.speeches-method
- hits-method: paramter cqp defaults to FALSE for hits-method, size
defaults to FALSE
- new parameter cqp for dispersion-method
- aggregation for dispersion-method when length(sAttribute) == 1
- bugfix for ngrams-method, sample code for the method
polmineR 0.6.1
- configure file removed to avoid unwanted bugs
polmineR 0.6.0
- this is the first version that passes all CRAN tests and that is
available via CRAN
- the ‘rcqp’ remains the interface to the CWB, but usage of rcqp
functions is wrapped into an new new CQI.rcqp (R6) class. CQI.perl and
CQI.cqpserver are introduced as alternative interfaces to prepare
portability to Windows systems
- code in the vignette and method examples will be executed
conditionally, if rcqp and the polmineR.sampleCorpus are available
- the polmineR.sampleCorpus package is available in a drat repo at
www.github.com/PolMine
- a series of bug fixes
polmineR 0.5.6
- slot tf renamed to stat, class is data.table now
- keyness_method moved to data.tables
polmineR 0.5.3
- renamed collocations to cooccurrences, seems more appropriate
polmineR 0.5.2
- multicore for term frequency counts (param for partition)
polmineR 0.5.0
- renamed xxxCluster to xxxBundle, bundle-superclass introduced
- slot label/labels renamed to name/names
- name/names-method instead of label/labels